[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kwin
Subject:    Re: GetFBConfigs
From:       Rivo Laks <rivolaks () hot ! ee>
Date:       2007-01-27 15:36:36
Message-ID: 200701271736.36582.rivolaks () hot ! ee
[Download RAW message or body]

Ühel kenal päeval (laupäev 27 jaanuar 2007) kirjutas Lubos Lunak:
> On Saturday 27 January 2007 14:09, Rivo Laks wrote:
> > Ühel kenal päeval (laupäev 27 jaanuar 2007) kirjutas Philip Falkner:
> > > Please test; I'm hoping this might get tfp working for you, Rivo.
> >
> > Yeah, TFP works now! Thanks a lot!
> > However, it often makes kwin crash when opening new window. Here's the
> > end of the debug log:
> >
> > kwin: Activation,
> > compared:KWinInternal::Client(0x83db148) :1671677425:1671675677:true
> > kwin: <unknown>(0x1800d61): 184
> > kwin: [void KWinInternal::SceneOpenGL::Window::createVertexGrid(int,
> > int)] 'Ark': Resized vertex grid from 0 quads (minreso: -1x-1) to 4 quads
> > (minreso: 0x0)
> > kwin: <unknown>(0x1800e88): BadPixmap (invalid Pixmap parameter)
> > kwin: <unknown>(0x1800e88): GLXBadDrawable
> > kwin: ERROR: [void
> > KWinInternal::SceneOpenGL::Window::findTextureTarget()] texture target is
> > 0, supports_npot: true
> > kwin:
> > /mnt/homes/home/rivo/build/KDE/kdebase/workspace/kwin/scene_opengl.cpp:92
> >5: void KWinInternal::SceneOpenGL::Window::findTextureTarget(): Assertion
> > `false' failed.
> > KCrash: crashing... crashRecursionCounter = 2
> >
> > Weird is that it only crashes in certain cases, not always. E.g. it
> > crashes when opening first Ark window, but if an Ark window is already
> > there and I open second (or third or forth, etc) one, it works fine. It
> > always crashes when opening Speedcrunch window.
> > I added the debug line which shows that target is 0. I have no idea why
> > it's 0, maybe you know better. Or maybe it's just a bug in NVidia
> > drivers...
>
>  Can you try completely removing the "GLX_TEXTURE_TARGET_EXT, tfp_target,"
> line from the attributes and check if that helps? It actually does not seem
> to be necessary here with either of the machine I have now at home, I'll
> need to check at work again what the problem was.

It doesn't make any difference here.

>  I also remember that after the npot patch the machine at work (GeForce FX
> 5500 IIRC) took noticeable time in bindTexture() in non-TFP modes, but here
> both GeForce2 MX 400 and GeForce4 4200 Go don't have this problem either
> (all using 9631 driver). Can you see it on your machine? Maybe there's
> something fishy with the work machine.

I don't see much of a difference between TFP and SHM here.
Your problem is probably that GeForce 5xxx doesn't really support npot 
textures. To support OpenGL 2.0, the driver emulates them for such cards 
which is possibly where your slowdown comes from.
It would likely be a good idea to just check for ARB_npot extension and omit 
the GL 2.0 check when checking if npot textures are supported.

Rivo
_______________________________________________
Kwin mailing list
Kwin@kde.org
https://mail.kde.org/mailman/listinfo/kwin

[prev in list] [next in list] [prev in thread] [next in thread]